Wednesday, May 30th, 2018

Lagging sectors and regions

Last week we did an exercise which compared our sector overweights with our regional equity overweights. The purpose was to determine which sectors have historically led their region higher when that region was also leading global equity markets. We use our normal asset allocation model and the local currency equity sector model to decide what to overweight. To get a measure of conviction we multiply the recommended regional overweight by the recommended sector overweight. Thus, if we are 50% overweight in the US and 70% overweight in Technology, the combined score would be 35%. We can calculate the average score for each sector for the periods when a region is overweight. We can also measure the frequency with which sectors have significantly positive scores and compare them with the frequency of significantly negative scores. This is potentially a very rich data set, because we can do the same in the other direction. This week we examine which sectors have historically been underweight when their region has also been underweight.

Calculating the numbers is easy but interpreting them is harder. This is because we are dealing with two different regimes. The first is when a region is underweight while global equity markets are rising; the second occurs when that region is leading global equity markets lower. The difference is one of causation. Is the sector the one that is responsible for the bear market like US Technology in 2000-01 and US Financials in 2008? Or, is it just a sector which is not doing well when the region is unable to participate in a bull market?

We could of course refine our sample to consider only periods when we are underweight in a region when we are also underweight equities relative to fixed income. This would help us identify which sectors actually cause bear markets. However, there are two problems with this approach. First, bear markets are less frequent than bull markets, so the samples are much smaller. Second, the results become quite sensitive to the definition of boundary conditions and to the choice of the non-equity risk-free asset. The aggregate scores and the net frequency may become a function of the severity of the downturn (2008) vs its duration (2000-03). The underlying message is that bear markets are more different from each other than bull markets, so this sort of exercise may not have much predictive power.

By contrast, part of the skill of portfolio construction is knowing which sectors to avoid when a region is out of favour, but portfolio limits require the fund manager to maintain a minimum level of exposure. It is useful in all market conditions not just bear markets. Full details are set out in the tables and charts which follow, but the main findings are as follows:

• Across all regions, the worst laggard by average score is Telecom, appearing in three regions with Small Caps, Financials and Utilities in two each. We get the same result looking at net frequency, apart from Small Caps.

• Looking at net frequency, the defensive sectors; Healthcare, Consumer Goods, Telecom and Utilities account for eight out of the top twelve underweight recommendations when the region is underweight and cyclical sectors (including Small Caps) account for ten of the top twelve overweight recommendations.

• In other words, being overweight a cyclical sector or underweight defensives in an underweight region probably does not compensate for being in the wrong region when equities are rising.

Leave a Reply

Full Blog Archive:

Other Recent Posts